Students get the better of AI grading algorithms / Humans + Tech - #45

+ Africa's youth don't find news on Facebook trustworthy + A supercomputer analyzed Covid-19 and has a new theory + Are we living in a computer simulation?


I hope you had a great week. Here are some interesting articles that I came across.

These students figured out their tests were graded by AI — and the easy way to cheat

Online learning due to Covid-19 has pushed many schools to outsource teaching and grading to education platforms. Many of these platforms use algorithms to grade students. The trouble is that they grade like the machines that they are. They look for keywords and certain patterns in the answers, not for actual understanding.

Dana Simmons, a history professor, realized that an algorithm was behind the grading when her son was extremely upset at getting a 50/100 on his first assignment. She gave him some pointers on how to game the system and within a day he figured out how to get a 100 on his tests without any actual understanding of the material [The Verge].

Simmons, for her part, is happy that Lazare has learned how to game an educational algorithm — it’s certainly a useful skill. But she also admits that his better grades don’t reflect a better understanding of his course material, and she worries that exploits like this could exacerbate inequalities between students. “He’s getting an A+ because his parents have graduate degrees and have an interest in tech,” she said. “Otherwise he would still be getting Fs. What does that tell you about... the digital divide in this online learning environment?”

Like algorithms that inherit and expose racial biases from the data humans train them with, educational algorithms are exposing the faults in our education system as well. Education for the current times needs a complete rethink and revamp. Grading algorithms are only a small part of the problem. In the information age, the way we teach and the way we learn has to change to adapt to the demands of our current societal and work requirements.

On a lighter note, where were these hacks to get a perfect 100 on tests when I was a kid? Kids have it so much easier these days :)

Facebook’s push to fix its fake news problem isn’t working in Africa either

4,200 young Africans from the ages of 18 to 24 were interviewed across 14 sub-Saharan African countries. More than half of them said they don’t find Facebook a trustworthy source of news [Quartz].

More than half of young Africans do not regard Facebook as a trustworthy source of news, claims a poll by the African Youth Survey, which was commissioned by the Ichikowitz Family Foundation and conducted by global polling firm, PSB Research. WhatsApp, the messaging app owned by Facebook, which is the dominant social media platform in Africa, is also deemed untrustworthy by half of the survey’s respondents. In contrast, only about a fifth of respondents had similar misgivings about Google as a source of news.

This makes me really happy. Disinformation and misinformation online is a huge problem, especially on Facebook, which prioritises profits over policies that benefit society. I’m really glad that most of the young Africans recognise that these platforms peddle a lot of fake information and are questioning the validity of the information they read. That is the best defence against disinformation and misinformation.

A supercomputer analyzed Covid-19 – and an interesting new theory has emerged

Scientists used the Summit supercomputer at Oak Ridge National Lab in Tennessee to analyse more than 40,000 genes from 17,000 genetic samples in an effort to better understand Covid-19 [Elemental].

Summit took a week to analyse all that data.

When Summit was done, researchers analyzed the results. It was, in the words of Dr. Daniel Jacobson, lead researcher and chief scientist for computational systems biology at Oak Ridge, a “eureka moment.” The computer had revealed a new theory about how Covid-19 impacts the body: the bradykinin hypothesis. The hypothesis provides a model that explains many aspects of Covid-19, including some of its most bizarre symptoms. It also suggests 10-plus potential treatments, many of which are already FDA approved. Jacobson’s group published their results in a paper in the journal eLife in early July.

One of my favourite uses of technology is when it saves lives or improves people’s lives by reducing pain and discomfort. We are still in the early stages of fully understanding Covid-19, and it may be years before we understand its long-term effects. However, any understanding we can gain to help people recover quickly in the near-term is welcome. I find it fascinating that a supercomputer can crunch the data and provide insights that we may have taken so much longer to realise on our own or may never even have thought of.

What are the odds we are living in a computer simulation?

This is an article from a few years ago, but I stumbled across it again and it’s something I often ponder about. I find hints in both spiritual teachings and science that this may be a possibility. It’s a very thought-provoking read [The New Yorker].

The argument is based on two premises, both of which can be disputed but neither of which are unreasonable. The first is that consciousness can be simulated in a computer, with logic gates standing in for the brain's synapses and neurotransmitters. (If self-awareness can arise in a lump of neurons, it seems likely that it can thrive in silicon, too.) The second is that advanced civilizations will have access to truly stupendous amounts of computing power. Bostrom speculates, for example, that, thousands of years from now, our space-travelling descendants might use nanomachines to transform moons or planets into giant “planetary computers.” It stands to reason that such an advanced civilization might use that computing power to run an “ancestor simulation”—essentially, a high-powered version of the video game “The Sims,” focussed on their evolutionary history. The creation of just one such simulated world might strike us as extraordinary, but Bostrom figures that thousands or even millions of ancestor simulations could be run by a single computer in the future. If that’s true, then simulated human consciousnesses could vastly outnumber non-simulated ones, in which case we are far more likely to be living inside a simulation right now than to be living outside of one.

Quote of the week

“He’s getting an A+ because his parents have graduate degrees and have an interest in tech, otherwise he would still be getting Fs. What does that tell you about... the digital divide in this online learning environment?”

— Dana Simmons, in the article, These students figured out their tests were graded by AI — and the easy way to cheat [The Verge]

I wish you a brilliant day ahead :)